
26 Jul
2022
26 Jul
'22
5:37 p.m.
my extended trained has for some reason 1 fewer vocab words than the tokenizer used for the model the past few days get to debug that!